2016年KDD IJCAI WWW ICDM 文章的个人统计

近两个星期，我大致浏览了2016年有关数据挖掘的几个会议文章的标题KDD,IJCAI,WWW,ICDM。通过阅读文献的标题与部分文章的摘要，大致了解文献的意图，并按照文献的工作内容分类文章，

浏览统计结果：

下面以关键字表格的形式列出总结结果(注：我只统计经常出现的一些关键词，并不是全部)：

表1：IJCAI-16文献统计（共551篇）
关键字	有关文献数	细分成分	关键字	有关文献数
planing	23	classical planning4	multi-agent	7
knowledge	23	Knowledge base 6 , knowledge graph 5 , knowledge acquisition 2	matrix factorization	7
representation	21		topic	6
recognition	20	表情，人类活动，行为，物体，感情，面部，群体活动	large scale	6
Graph	19		track	5
games 游戏	19		social network	5
feature	19	feature learning 7，feature selection 6	sampling	5
semantic	18		regression	5
Neural Network	17		Natural Language	5
Clustering	17		multi-task	5
query	15	query answering 5	deep learning	5
detection	14	community detection 2 ，anomaly detection 1	retrieval	4
Recommendation	13		ranking	4
classification	12		person re-identification	4
robot	11		coding	4
Bayesian	11		time series	3
tree	10	Mote-Clarlo tree search 4	probabilistic matrix	3
text	8		random forest	2
hashing	8		dictionary learning	2
filtering	8	collaborative filtering 7	advertisements	2
multi-view	7

表2:WWW-16(共100篇左右)
关键字	相关文献数	关键字	相关文献数	关键字	相关文献数
web（web search, web cookies, web shells, web tables, web tracking, web queries ,web application）	17	topic	4	knowledge bases	2
recommendation	10	search engines	4	clustering	1
detection	9	advertisement	3	representation	1
query	7	filter	3	Bayesian	2
mobile	6	large-scale	3	documents	3
text	2	bayain	2	graph	1
semantic	4	track	2

表3：KDD-16（共208篇）
关键字	相关文献数	关键字	相关文献数	关键字	相关文献数
Graph	12	document	4	hashing	2
recommendation	10	neural network	4	multi-task	2
clustering	9	social network	4	multi-view	2
optimization	9	text	4	query	2
feature	8	classification	3	regression	2
large scale	8	sampling	3	semantic	2
feature	7	topic	3	community detection	1
anomaly	6	attributed network	2	matrix completion	1
rank	5	filtering	2	pagerank	1

表4:ICDM-16(共201篇 含Demonstrations)
关键字	有关文献数	关键字	有关文献数	关键字	有关文献数
graph（graph Decomposition ,   supergraph serach , communyity Detection, knowledge graph）	27	feature	4	tree	3
query	14	semantic	4	advertisement	3
clustering	13	text	4	filtering	2
KNN	7	hashing	3	topic	2
recommendation	6	large-scale	3	factorization	1
classification	5	neural network	3	multi-task	1
group search	5	sampling	3	mults-view	1
social network	5

有关发现与理解：

从这些表格中，我发现大家主要在做的可以分为几个大方面：
1.features  2.clustering 3.classification 4.detection 5.representation 6recognition 7.planning(这个我现在还不是很理解是什么)

当然 还有:

8.filtering 9.query (这个也不是很理解是什么) 10.regression 11 ranking 12 sampling 13 track

有关应用方面:

做的最多的还是1.recommendation  广告推荐 新闻推荐 ;

其次还有 2.social network  3. topic 主题分析；4 search 

当然数据挖掘的应用面很广，我在阅读时记录下来的应用范围就有如下：

出租车 广告业 新闻业  商务业 医疗 广播业 交通 投资贷款 城市规划 生物基因 经济

电子商务  诉讼 社交网络 物理  学校评估 数据中心储存盘替换 警务 法律  体育（足球）航空

邮件处理  招聘 抓小偷 房地产 火灾 说唱歌词 猎头人才检测 工作推荐 在线采购 病毒检测 

甚至政党领导人选举等等。

从统计中看，数据挖掘中用到的数据结构主要是：graph与tree，（感觉graph 很火的样子）。至于用到的方法 neural network 还是很热门，其次bayesian也有很多人用。hashing 也有不少人在讨论，矩阵操作也是有很多。

还有一些有意思的发现：

1.每个会议都有一些文章在讨论’large-scale’数据或者相关的问题。

2.multi-task  multi-view multi-agent  在每个会议中也都会出现

问题与解决

在阅读的过程中也遇到了一些问题，看不懂在标题意思，甚至词语意思都不懂，（这些词语我都记录了下来）。对于感兴趣的文章我也记录下来，留有以后学习理解。

问题如下：

1.大尺寸数据处理越来越热？

2.连百度hr都可以发kDD

3.有这么多做航空公司的为什么没有做铁路公司的？ 有多人做交通网络 公路交通网 包涵了铁路

不懂的词汇（留有以后懂了之后再补上解释）：

Maximum Weight Clique Problem?

ASP？An ASP Semantics for Default Reasoning with Constraints

470

SMT?

Proving the Incompatibility of Efficiency and Strategyproofness via SMT Solving

 How to Build Your Network? A Structural Analysis

Item Recommendation for Emerging Online Businesses

NP hard ？

bilief? embedding?

Out-of-Sample Data?

Distributed?

Heuristics ? 启发?

Directional Statistics?

Plan Recognition as Planning Revisited? planning?

logics? 逻辑？

Hierarchical？Hierarchical model Hierarchical planning

*-based?

query?

Temporal Graph?

Graph Streams?

join?

KNN?

Embedding？

time-seris?  时间序列分析研究的时间序列？

感兴趣的文章：

KDD: http://www.kdd.org/kdd2016/program/accepted-papers

A Real Linear and Parallel Multiple Longest Common Subsequences (MLCS) Algorithm

Taxi Driving Behavior Analysis in Latent Vehicle-to-Vehicle Networks: A Social Influence Perspective

Predicting Socio-Economic Indicators using News Events

Minimizing Legal Exposure for High-Tech Companies through Collaborative Filtering Methods

Identifying Decision Makers from Professional Social Networks

Catch Me If You Can: Detecting Pickpocket Suspects from Large-Scale Transit Records

Ranking Universities Based on Career Outcomes of Graduates

DopeLearning: A Computational Approach to Rap Lyrics Generation

Analyzing Volleyball Match Data from the 2014 World Championships Using Machine Learning Techniques

Collective Evolution Inference in Heterogeneous Information Networks  

Lexis: An Optimization Framework for Discovering the Hierarchical Structure of Sequential Data

IJCAI:http://ijcai-16.org/index.php/welcome/view/accepted_papers

Multi-view Exclusive Unsupervised Dimension Reduction for Video-based Facial Expression Recognition

Dimensionally Guided Synthesis of Mathematical Word Problems

Truncating Shortest Path Search for Efficient Map-matching

Hidden Parameter Markov Decision Processes: A Semiparametric Regression Approach for Discovering Latent Task Parametrizations

Semi-Supervised Multimodal Deep Learning for RGB-D Object Recognition

A Neural Network for Document Summarization

Improving Top-N Recommendation with Heterogeneous Losses

Transductive Optimization of Top k Precision

Chinese Song Iambics Generation with Neural Attention-based Model

Derivative-Free Optimization of High-Dimensional Non-Convex Functions by Sequential Random Embeddings

Fear and Hope Emerge from Anticipation in Model-Based Reinforcement Learning

 Driver Frustration Detection From Audio and Video in the Wild

Moving in a Crowd: Safe and Efficient Navigation among Heterogeneous Agents

Predictive models of malicious behavior in human negotiations

 A Framework for Recommending Relevant and Diverse Items

Matrix Factorization+ for Movie Recommendation

ICDM:http://www.icde2016.fi/papers.php#tabular1

Context-Aware Advertisement Recommendation for High-Speed Social News Feeding

A Novel Fast and Memory Efficient Parallel MLCS Algorithm for Longer and Large-Scale Sequences Alignments

A Framework for Enabling User Preference Profiling through Wi-Fi Logs

WWW:http://www2016.ca/2-home/72-accepted-papers.html

Immersive Recommendation: News and Event Recommendations Using Personal Digital Traces

Exploiting Green Energy to Reduce Operational Costs of Multi-Center Web Search Engines 

Scaling up Dynamic Topic Models (jun zhu)

The Lifecycle and Cascade of Social Messaging Groups(jie tang)

Impact, Characteristics, and Detection of Wikipedia Hoaxes